[Labs-l] ToolLabs back up

Wed Feb 4 19:30:00 UTC 2015

>Well, I think this is the problem right here. If something must have
>"24/7 uptime", it should be in production, not labs.
Not really, that would be the case if it could not tolerate outages of any
sort, 'ideally' we would like our tools to be 24/7. Now, can the tools
tolerate a full-day outage? They sure can, the issue is that - as of the
last quarter- we have had several outages not just
one. And some of them have lasted over a day and that is when things start
to be problematic for our users.

We get that there had been a bunch of events related to db, memory, virt
machines that had to get fixed but to be clear we are not asking for
production-type services (we also own several of those and they are
deployed on our prod stack) we just would like to have a slightly better
uptime in labs for our tools.

Looks like a bunch of things related to firewalls, IP rules, db replication
and several others have been fixed as of late, we can take it up from here
and see how does the uptime do going forward.

Thanks,

Nuria

On Wed, Feb 4, 2015 at 11:17 AM, Legoktm <legoktm.wikipedia at gmail.com>
wrote:

> On 02/02/2015 07:04 PM, Dan Andreescu wrote:
> > Since a lot
> > of people depend on those tools having pretty much 24/7 uptime
>
> Well, I think this is the problem right here. If something must have
> "24/7 uptime", it should be in production, not labs.
>
> -- Legoktm
>
> _______________________________________________
> Labs-l mailing list
> Labs-l at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/labs-l
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.wikimedia.org/pipermail/labs-l/attachments/20150204/350b6dc4/attachment-0001.html>